1 Revision # 2 – January 31st, 2014
2
3 Determination of the concentration of asbestos minerals in highly
4 contaminated mine tailings. The example of the abandoned mine waste
5 of Crètaz and Èmarese (Valle d’Aosta, Italy).
6
1, 1 1 7 ALESSANDRO F. GUALTIERI, * SIMONE POLLASTRI, NICOLA BURSI GANDOLFI,
1 2 3 8 FRANCESCO RONCHETTI, CARLO ALBONICO, ALESSANDRO CAVALLO, GIOVANNA
4 4 5 9 ZANETTI, PAOLA MARINI, AND ORIETTA SALA
10 1Dipartimento di Scienze Chimiche e Geologiche, Università di Modena e Reggio Emilia, I-41121 Modena,
11 Italy
12 2A.R.P.A. Valle d’Aosta, I-11020 Saint Christophe Aosta,Italy
13 3Dipartimento di Scienze dell'ambiente e del territorio e di Scienze della terra, Università di Milano-
14 Bicocca, I-20126 Milano, Italy
15 4Dipartimento di Ingegneria delle Materie Prime DIATI - Politecnico di Torino, I-10129 Torino, Italy
16 5A.R.P.A. Emilia Romagna, Sezione Provinciale di Reggio Emilia
17 I-42122, Reggio Emilia, Italy
18
19 Submitted to The American Mineralogist
20 *Corresponding author.
21 Phone: +39 059 2055810; Fax: +39 059 2055887
22 E-mail address: [email protected]
23 ______
1
24 ABSTRACT
25 For the first time, this work reports concentration maps of asbestos minerals in
26 contaminated mine tailings drawn using the results of Rietveld quantitative phase
27 analysis (QPA). The investigates sites are located in the Valle d’Aosta region (Italy):
28 Crètaz, the most important Italian magnetite mine, active until 1979 and Emarèse, one the
29 most important chrysotile asbestos mines in Italy, active until 1968. The results of the
30 study permit to draw the spatial distribution of the asbestos (chrysotile and tremolite in
31 this specific case) concentration, useful to plan reclamation of the sites, with priority
32 given to the areas with the highest asbestos concentration. Because of the complexity of
33 the mineral assemblage which includes, among the others, antigorite, chlorite, talc, and
34 tremolite, the concentration of chrysotile was cross-checked using different experimental
35 techniques such as X-Ray powder diffraction (XRPD), Fourier Transform Infra-Red
36 (FTIR) spectroscopy, scanning electron microscopy (SEM), Polarized light Optical
37 Microscopy (PCOM), and differential thermal analysis (DTA). The accuracy of the
38 results was validated by analyzing standard samples with known concentrations of
39 chrysotile and tremolite. The comparison allowed to point out the advantages and
40 disadvantages of each experimental method.
41 At Crètaz, chrysotile ranges from 4.4 to 22.8 wt% and tremolite from 1.0 to 10.3 wt%
42 whereas at Emarèse the concentration of chrysotile varies from 3.3 to 39.5 wt% and
43 tremolite from 5.9 to 12.4 wt%. Antigorite and chlorite are the major accompanying
44 phases with variable amounts of other accessory minerals including magnetite,
45 carbonates, talc, olivine, pyroxene, talc, and brucite. The results of our study are of key
46 importance for the local environmental policies as the knowledge of the spatial
2
47 distribution of the asbestos concentration allows to plan a detailed reclamation agenda of
48 the contaminated sites. The spots with the highest surface contamination of both
49 chrysotile and tremolite were identified and classified as priority area in the reclamation
50 plan.
51 Keywords: Chrysotile, tremolite, serpentine, mine tailings, quantitative determination
52
53 INTRODUCTION
54 Commercial asbestos minerals are classified into two groups: serpentine and
55 amphibole asbestos, both sharing the same fibrous-asbestiform crystal habit but different
56 structural arrangements at the molecular scale (Bailey 1988). The fibrous-asbestiform
57 variety of serpentine is chrysotile (white asbestos) with ideal chemical formula
58 Mg3(OH)4Si2O5. The amphibole asbestos family includes five minerals: actinolite,
59 amosite (fibrous variety of grunerite, brown asbestos), anthophyllite, crocidolite (fibrous
60 variety of riebeckite, blue asbestos), and tremolite. Amphiboles’ have general formula
2+ 2+ 2+ 61 AB2C5T8O22W2 with A = , Ca, K, Li, Na; B = Ca, Fe , Li, Mg, Mn , Na; C = Fe ,
62 Fe3+, Li, Mg, Mn2+, Mn3+, Ti4+; T = Al, Si, Ti4+; W = Cl, F, O2-, (OH). The ideal
63 (approximated) formulas of the five asbestos species are: actinolite A = , B = Ca, C =
64 (Fe2+, Mg), T = Si; W = (OH); amosite A = , B = C = Fe2+, T = Si; W = (OH);
65 anthophyllite A = , B = C = Mg, T = Si; W = (OH); crocidolite A = , B = Na, C =
66 (Fe2+, Fe3+, Mg), T = Si; W = (OH); tremolite A = , B = Ca, C = Mg, T = Si; W = (OH).
67 The six asbestos minerals exhibit outstanding chemical-physical and technological
68 properties exploited for various industrial applications. Chrysotile has been the most
69 commonly used form of asbestos. Regrettably, the unique fibrous-asbestiform crystal
3
70 habit and surface activity, responsible for the excellent technological properties, also
71 seem to be the cause of asbestos minerals’ potential health hazard. The history of the
72 epidemiological reports of asbestos related diseases is well described in Skinner et al.
73 (1988). Although many epidemiological studies provided evidence that amphibole
74 asbestos minerals are more hazardous than chrysotile (Hodgson and Darnton 2000), all
75 six asbestos mineral species are assumed to be harmful to human health. Exposure
76 through inhalation of asbestos minerals may provoke lung diseases (Skinner et al. 1988;
77 Dilek and Newcomb 2003) and during the 1980s, asbestos minerals were declared proven
78 human carcinogens by the US Environmental Protection Agency, the International
79 Agency for Research on cancer (IARC) of the World Health Organization and the
80 National Toxicology Program (Nicholson 1986; IARC 1977; Collegium Ramazzini
81 2010). Later, many countries worldwide banned or restricted the use of asbestos
82 containing materials (ACMs). In the countries where asbestos minerals are banned,
83 ACMs have been progressively removed from the environment to minimize exposure of
84 the population (Gualtieri 2012). Despite the huge efforts to entirely remove asbestos from
85 the living environment, some issues are still matter of concern. One of these regards the
86 correct evaluation of the health hazard associated with abandoned asbestos mines
87 (especially dumps and tailings) and natural occurring asbestos (NOA). Mining of
88 asbestos and asbestos associated minerals (e.g., iron oxides) generated vast amounts of
89 residue material, chemically not very different from the original rock (Meyer 1980). An
90 important difference is the fineness of the residue material composing the dumps, making
91 it more prone to weathering or erosion by wind, with subsequent release of fibers in air or
92 in the percolating surface hydrological system. With this premise, Viti et al. (2011)
4
93 suggested that the correct evaluation of the hazard associated with serpentinite outcrops
94 (and asbestos containing mine tailings) would require a careful geo-statistical and geo-
95 mechanical investigation of the fracturing/vein system, followed by careful qualitative
96 and quantitative determinations. Unfortunately, the accomplishment of accurate and
97 precise quantitative figures of the actual asbestos minerals content is made difficult by
98 two major factors: (1) asbestos bearing serpentinite textures typically consist of fine-to-
99 ultrafine intergrowths of fibrous and non-fibrous minerals, often difficult to identify by
100 conventional methods, such as X-ray diffraction or microanalytical approaches. To this
101 aim, DTA seems to be an effective and promising tool of analysis although only a case
102 study has been reported so far (Viti et al. 2011) and further experimental evidence is
103 needed; (2) asbestos and asbestos containing mine dumps and tailings invariably contain
104 high concentrations of asbestos minerals (generally higher than 1wt%). In countries like
105 Italy, regulated experimental techniques for massive materials (X-Ray Powder
106 Diffraction - XRPD, Fourier Transform Infra-red spectroscopy - FTIR, Scanning Electron
107 Microscopy - SEM, and Phase Contrast Optical Microscopy - PCOM) are applied to
108 determine whether asbestos concentration is higher than 0.1 wt% (Italian D.Lgs
109 10/03/2010 nr. 205 All. D). On the same line, Airborne Toxic Control Measure (ATCM)
110 restricted in the USA the use of serpentine and ultramafic rock aggregates for surface
111 applications to materials containing less than 0.25 wt% asbestos determined according to
112 the Californian Air Resources Board method (CARB method 435, 1991). This scenario
113 reveals that accuracy of these quantitative methods have not been tested for highly
114 contaminated massive materials.
115 This work was prompted by the actual need to map the concentration of asbestos
5
116 minerals in abandoned asbestos or asbestos containing open mines with dumps and
117 tailings possibly containing very high concentrations of asbestos fibers (well above 1
118 wt% ). Because the costly reclamation of these sites usually conflicts with the tight
119 budgets of the regional/national funds allocated to such expense items, a step by step long
120 term plan must be proposed with priority given to the reclamation of the areas with the
121 highest asbestos concentration. Therefore, the knowledge of the spatial distribution of the
122 asbestos concentration in such sites is relevant from an economic standpoint as it may
123 permit to prioritize the most hazardous areas within a long term reclamation plan. For the
124 first time, we report the mapping of the concentration of asbestos minerals (chrysotile and
125 tremolite) in abandoned sites with highly contaminated mine tailings. The sites under
126 investigation are located in the Valle d’Aosta region in Italy: Crètaz is situated near the
127 town of Cogne where the most important Italian magnetite mine was active until 1979;
128 Emarèse was one the most important chrysotile asbestos mines in Italy, now abandoned
129 since 1968.
130 The determination of the chrysotile concentration was accomplished using all the
131 experimental techniques permitted by current Italian regulations for massive materials
132 XRPD, FTIR, SEM, and PCOM. Quantitative PCOM results were integrated by
133 qualitative observations in polarized light (PLOM) and with chromatic dispersion (CD).
134 Although the Italian D.M. 09/06/1994 suggests the use of SEM when the estimated
135 asbestos concentration is lower than 1wt% and MOCF, XRPD and FTIR when it is
136 higher than 1wt%, there is actually no indications on the upper concentration limits of
137 application of SEM. Hence, the analysis of samples with very high concentration of
138 asbestos offers a proper case study. The accuracy of the results was validated by the
6
139 analysis of especially prepared standard samples with known concentration of chrysotile
140 and tremolite and cross-checked by the application of the Differential Thermal Analysis
141 (DTA) and the Rietveld methods. The comparison allowed to point out the advantages
142 and disadvantages of each experimental method.
143
144 GEOLOGY OF THE INVESTIGATED AREA
145 The main structural lines of the Aosta valley (Fig. 1) are aligned N-W to S-E. They
146 represent a natural cross-section through the structural units of the Italian Alps where the
147 ancient European and African continental margins are divided by relict portions of the
148 oceanic floor. The margin of the European basement emerges in correspondence to the
149 crystalline massif externally represented by the Monte Bianco (Elvetico - Ultraelvetico
150 Domain) and internally by the Gran San Bernardo – Ruitor, the Monte Rosa and the Gran
151 Paradiso (Pennidico Domain). The ancient portion of the Piedmont ocean (the Piedmont
152 Zone with Calc-schists and Green Stones) occurs in the middle stretch of the Valley,
153 spanning the inner margin of the crystalline massifs. The area of the lowest valley is
154 mainly composed by the Pennidic and Austroalpine domains, characterized by folded
155 structures and marked by metamorphism in association with large thrust of the Piedmont
156 Zone of Calc-schists with Green Stones (ophiolites). The mining area of Cogne is found
157 in that formation and emerges in the localities of Aymavilles, Urtier river, and Grauson
158 river (Elter 1987). The rocks of that unit are known to host mineralizations of Cr, Pt, Ni
159 and Fe and various phases such as asbestos minerals and talc. Iron is mostly hosted in
160 magnetite, one of the main constituents of the serpentinite accessories, with the genesis
161 attributed to the processes of serpentinization (hydrothermal alteration of the original
7
162 mantle peridotites) which began probably in the oceanic environment and continued
163 during the Alpine orogeny (Dal Piaz 1971). The magnetite mines of Cogne, and in
164 particular the site of Crètaz, are located inside the area of the Piedmont Calc-schists and
165 Green Stones (Diella et al. 1994). The ore mining of Cogne was both underground and
166 open (Di Colbertaldo et. al. 1967) and the iron ore has been industrially exploited from
167 1900 to 1979.
168 The asbestos Émarèse mine is located inside the Piedmont Zone with Calc-schists and
169 Greenstones (ophiolites). The metabasites which occur at the Émarèse area are mainly
170 composed of serpentinites locally mineralized to antigorite and chrysotile. The latter
171 sometimes occurs in very long fibers that can reach a length of 1 m. Serpentines are
172 accompanied by carbonates, magnetite and talc. The mining activity regarded an open pit
173 and several wells and tunnels (Cavinato 1964). The Émarèse mine is abandoned since
174 1968.
175
176 EXPERIMENTAL PROCEDURE
177 Materials sampling and sample preparation
178 A number of representative samples were collected in the areas of Crètaz and
179 Emarèse. Given the extremely hazardous situation due to the proximity of the
180 contaminated site with the touristic town of Cogne, both surface and core sampling was
181 conducted at Crètaz whereas only surface sampling was conducted at Emarèse. As far as
182 the surface sampling is concerned, the materials were collected under a 25-30 mm thick
183 surface layer of soil manually removed with the aid of a shovel. All the operations were
184 conducted using protective equipments. Samples were collected down to a depth of about
8
185 60 mm with the aid of a manual corer. A representative amount of material, depending on
186 the average grain size, was sampled and manually divided with the aid of a Retsch
187 stainless steel sample splitter (maximum capacity = 16 l; maximum grain size = 50 mm)
188 to obtain about 2 Kg of homogeneous final raw product that was finally stored inside
189 glass bottles and sealed. At Crètaz, core sampling was possible using a rotary core drill
190 with a penetration length of about 20 m. Raw core samples were divided and
191 homogenized using the same procedure described above for surface samples. All the
192 surface and core collection spots for both localities are reported in Figure 1. The figure
193 legend also specifies the sampling mode.
194 In the lab, using protective equipments and under extractor fan, raw samples were
195 quartered to obtained about 50 g of homogeneous representative material and
196 subsequently dried at 105 °C for 2 h. Powders were obtained by mechanical milling using
197 Retsch mm 200 SiC jars, with a capacity of 10 ml each, for 10 min at a speed of 25
198 oscillations/s. Further fine manual powdering was conducted using an agate mortar for 5
199 min.
200 Two standard samples (labeled STD1 and STD2) with known contents of chrysotile
201 and tremolite asbestos and mineralogical composition similar to that of the natural
202 samples were prepared as weighed mixtures of artificial tailings to assess the accuracy of
203 the quantitative determination. The compositions of the standard samples (wt%) were:
204 STD1 (chrysotile 5.11, tremolite 5.11, talc 5.11, clinochlore 5.11, antigorite 8.07, calcite
205 40.86, dolomite 10.2, magnesite 4.09, albite 16.34); STD2 (chrysotile 10.44, tremolite
206 1.04, talc 5.22, clinochlore 5.22, antigorite 16.49, calcite 36.53, dolomite 10.44,
207 magnesite 4.18, albite 10.44). Specimens of the mineral phases talc, clinochlore, calcite,
9
208 dolomite, magnesite, and albite were taken from the mineral collection of the MUSEO
209 GEMMA 1786, Dipartimento di Scienze Chimiche e Geologiche of the University of
210 Modena e Reggio Emilia (Modena, Italy).
211 The asbestos samples used to prepare the weighed mixtures of artificial tailings are the
212 NIST chrysotile standard SRM1866a and the NIST tremolite sample SRM 1867a, the
213 highest purity available. Bunches of fibers were ultrasonicated and selected under optical
214 microscope. The antigorite sample is from a pale-green, splintery vein occurring in a
215 massive serpentinite from Elba Island (Italy): sample ATG18 in Viti and Mellini (1996).
216 The sample predominantly consists of antigorite lamellae with an average composition of
2+ 3+ 217 Mg2.62Fe 0.16Fe 0.03(OH)4Al0.01Si2O5 (normalized to 2 apfu in the tetrahedral site; Loss
218 of Ignition (L.O.I.) of 11.9 wt%) and super-periodicity of 48.8 Å; interstitial chrysotile
219 occurs among antigorite lamellae. The total chrysotile content is 22.0(9) wt% (Viti et al.
220 2011). Obviously, the chrysotile content of antigorite was considered during the
221 preparation of the mixture. Sample purity of all the selected specimens was checked by
222 XRPD. Prior to the weighing procedure, all the specimens were dried at 105 °C for 2 h
223 and reduced to powders by manual grinding in an agate mortar using acetone as solvent.
224 After weighing the various components, the coarse powders were homogenized and
225 powdered in an agate mortar for 5 min.
226
227 Analytical methods
228 Sample selection
229 Two natural samples (C2 and E1) and the two standards STD1 and STD2 were
230 selected for the determination of the concentration of chrysotile and tremolite (when
10
231 possible) asbestos, following the experimental methods considered in the Italian D.M.
232 09/06/1994, differential thermal analysis (DTA) and X-ray powder diffraction (XRPD)
233 with the Rietveld method. All the other samples collected during the surface and core
234 sampling at Crètaz and Èmarese were analyzed with the Rietveld method to determine
235 both the concentration of chrysotile and tremolite asbestos.
236 Optical microscopy
237 As far as the optical microscopy observations are concerned, prior to the qualitative
238 analysis, the samples underwent a chemical attack using a very diluted HCl solution to
239 remove the carbonate fraction and eventually hydrated carbonate alteration products and
240 release the fibers from the matrix. About 0.2 mg of the powder was then suspended in a
241 drop of acetone on a slide, put in a bottle, and inserted in an ultrasonic bath for 15 min. A
242 liquid of proper refraction index is added to the powder on the slide for the polarized light
243 optical microscopy (PCOM) observations. The following refractive indices were used for
244 the identification of the asbestos fibres in the samples: 1.550 for chrysotile; 1.610-1.615
245 for tremolite; 1.620 for anthophyllite, and 1.640 for actinolite. The microscopes used for
246 the qualitative observations were a Leica DMLP and Leica DIALUX 20.
247 The samples for the quantitative analyses were prepared by suspending 0.1 mg of
248 powder in 10 ml of distilled water. The suspension was ultrasonicated for 15 min. The
249 suspension was filtered using a 47 mm cellulose nitrate filter to obtain an homogeneous
250 dispersion. The filter was subsequently diaphanized for the PCOM observation. The
251 observations were performed at 500x over a 1.57 mm2 area of the membrane (0.5% of the
252 total membrane area of 152 mm2). Each regulated (length > 5 μm and aspect ratio of 3)
253 fiber of both chrysotile and tremolite asbestos observed on the membrane was reported
11
254 and summed up to calculate the total volume. Assuming theoretical density values of
255 crysotile and tremolite fibers of 2.6 g/cm3 and 3.0 g/cm3, respectively, it was possible to
256 calculate the total weight of both asbestos species counted on the membrane. This value
257 was then rescaled to the entire powdered area of the membrane and therefore to the
258 amount (0.1 mg) of powder on it.
259 Electron microscopy
260 The scanning electron microscopy (SEM) determinations were performed following
261 the method described in the Italian legislation (Italian Ministry of Health, 1994), which is
262 very similar to the International Standard Organization (ISO) Method 14966 (ISO, 2002).
263 The fiber-definition criteria are those cited above and are compliant with the ISO 14966
264 and WHO standard (WHO 1997; ISO 2002). An amount of 10 mg of each powder sample
265 was suspended in 200 ml of deionised and micro-filtered water with 0.1 vol.% surfactant
266 additive (dioctyl sodium sulfoccinate, C20H37NaO7S, CAS nr. 577-11-7), and
267 ultrasonicated to facilitate the particle dispersion. A volume of 3 x 2.5 ml of this
268 suspension was collected at different levels and put into a filtering system, allowing
269 random deposition of the particles on polycarbonate filters (Osmonics 25 mm diameter,
270 0.6 µm porosity). Filters were dried in an oven (at 55 °C), weighed using a high-precision
271 balance and coated with graphite. The samples were analyzed with a Vega TS Tescan
272 5163 XM SEM (LAB1, the University of Milan-Bicocca), in combination with an
273 energy-dispersive X-ray spectrometer (EDAX Genesis 400), operating at 20 kV and 200
274 pA, at a working distance of about 20 mm. For each sample, 1 mm2 of the filter surface
275 was investigated, working at 2000–6000 magnification. The technique is based on point-
276 counting statistics, reporting the occurrence, number, dimension and chemical
12
277 composition of asbestos fibers in each measured spot. The volume of the single fiber is
278 approximated to that of a cylinder, and the weight is calculated assuming an average
279 density of 2.6 g/cm3 for chrysotile and 3.0 g/cm3 for all amphibole species. The fiber-
280 concentration C (in ppm) was then calculated as follows:
6 281 C = [A•(wc + wa)/n•a•W]•10
2 282 with A = filter surface (mm ); wc = total weight of counted chrysotile fibers (mg); wa =
283 total weight of counted amphibole fibers (mg); n = number of analyzed spots; a = spot
284 surface (mm2); W = weight of the sample on the filter (mg).
285 The samples (following the protocol of LAB1) were also prepared separately and
286 analyzed in another laboratory (LAB2, A.R.P.A. Aosta). An amount of 10 mg of each
287 powder was suspended in 200 ml of deionised and micro-filtered water with 0.2 vol.%
288 surfactant additive (sodium lauryl sulfate, CH3(CH2)11OSO3Na, CAS nr. 151-21-3), and
289 ultrasonically treated. All the other experimental conditions were identical to those
290 reported above. The instrument used for the observations was a Zeiss EVO MA10 with a
291 LaB6 cathode, in combination with an EDAX system, operating at 20 kV and 200 pA,
292 with 500–16000 magnification, at a working distance of 12 mm.
293 Infrared spectroscopy
294 For the Fourier transform Infrared (FTIR) spectroscopy measurements, pellets were
295 prepared using 198 mg of dried KBr powder and 2 mg of dried sample powder, with a
296 final load of 12 t for 12 s. A blank standard sample composed of KBr powder was also
297 prepared. Prior to the analysis, a calibration curve with increasing amounts of chrysotile
298 was prepared using the same experimental conditions to apply the Linear Calibration
299 Curve Method (see a through description of the method applied to FTIR in De Stefano et
13
300 al. 2012). All the spectra were collected from 4000 to 600 cm-1 at 4 cm-1 resolution using
301 a Bruker VECTOR 22 spectrometer with a Globar source, aperture 3 mm, 32 scans,
302 velocity 10 kHz, detector DTGS with a window of KBr (8000–400 cm-1), Mertz
303 correction phase, apodization function 3-term Blackman-Harris. Data were collected and
304 analyzed using the Bruker OPUS 6.5 software. The intensity of the absorption band at
305 3697 cm-1, the stretching vibration of chrysotile non-hydrogen-bonded OH- (Farmer
306 1974), was measured and plotted versus the added chrysotile weight percent. It is
307 assumed that the detection limit is 0.01 wt% (Foresti et al. 2003). This procedure is rather
308 straightforward when no interference phases such as serpentine polymorphs and chlorite
309 are present (see for example, Balducci and Valerio 1986). The presence in the
310 investigated samples of the layer silicates antigorite and chlorite, which may overlap their
311 absorption bands to that of chrysotile, requires a careful discrimination to obtain the
312 intensity of the chrysotile band to plot vs. chrysotile concentration. For the investigated
313 samples with unknown concentration of chrysotile, the intensity of the absorption band
314 was measured as closest as possible to the ideal value of 3697 cm-1 (Farmer 1974), using
315 the “R” type integration process available in the software (the value of the intensity
316 measured above the base line taken between 3720 e 3661 cm-1). The resulting values
317 were then used to extract the chrysotile concentraion from the calibration curve.
318 X-ray powder diffraction
319 The XRPD quantitative analysis of chrysotile asbestos has been long conducted using
320 the external standard method (Klug and Alexander 1974) where the peak
321 intensity/integrated area of the major diffraction peak of an analyte (001 peak for
322 chrysotile) is usually plotted vs. its concentration to build a calibration curve. The
14
323 concentration of the analyte in an unknown sample requires the determination of the
324 correct unbiased intensity/integrated area of the major diffraction peak. Puledda and
325 Marconi (1990) developed a method used for chrysotile legitimated by the Italian D.M.
326 09/06/1994 which is based on the use of a silver membrane filter. The method also
327 proposes an appropriate value for the deposition area of the sample on filter, which
328 optimizes the diffraction response of the analyte. To apply the method described above,
329 the powder patterns of the samples C2, E1 and the two standards STD1 and STD2 were
330 collected using a Bragg–Brentano θ-θ diffractometer (PANalytical X’Pert Pro), Cu Kα
331 radiation, 40 kV and 40 mA) equipped with a curved graphite monochromator and a gas
332 proportional detector. The divergence, receiving and anti-scattering slits were of 1°, 0.1
333 mm, and 1°, respectively. The scans were collected with a step scan of 0.02 °2θ, and 5
334 s/step in the range 4-64 °2θ. The data analysis was performed using the PANalytical
335 HighScore Plus software version 2.2c.
336 The mineralogical quantitative phase analyses (QPA) of all the Crètaz and Èmarese
337 samples and the two standard samples STD1 and STD2 using the Rietveld method
338 (Rietveld 1969) were performed through X-ray powder diffraction (XRPD). Data were
339 collected using a Bragg–Brentano θ-2θ diffractometer (Bruker D8 Advance, Cu Kα
340 radiation, 40 kV and 40 mA) equipped with a Göbel mirror on the incident beam and a
341 solid state detector. The width of the divergence, receiving and anti-scattering slits were
342 of 0.6, 0.1, and 0.2 mm, respectively. The scans were collected with a step scan of 0.02
343 °2θ, and 20 s/step in the range 4-65 °2θ. QPAs were performed following the refinement
344 strategies described in Gualtieri (2000). Refinements were accomplished with the GSAS
345 (Larson and Von Dreele 1999) package and its graphical interface EXPGUI (Toby 2001).
15
346 Whenever the samples showed the presence of more than the maximum number of 9
347 phases available to the program, the procedure described by Winbur et al. (2000) was
348 used. Although samples were prepared with the side loading technique to minimize a
349 priori preferred orientation of crystallites, residual preferred orientation effects were
350 modelled with the March model (Dollase 1986). The following structural models were
351 used: albite (Downs et al. 1994) and plagioclase (Wenk et al. 1980), antigorite (Uehara
352 1998), calcite, dolomite, quartz (Le Saoût et al. 2011), chrysotile (Falini et al. 2004),
353 chlinochlore (Zanazzi et al. 2009), illite (Birle 1968), magnetite (Haavik et al. 2000),
354 microcline (Ribbe 1979) and orthoclase (Prince et al. 1973), talc (Perdikatsis and Burzlaff
355 1981), and tremolite (Ungaretti and Oberti 2000).
356 Thermal analysis
357 The collection of the thermal analysis curves thermogravimetry (TG), differential
358 thermogravimetry (DTG), and DTA of the C2, E1, and the two standard samples STD1
359 and STD2 samples were performed using a simultaneous differential thermal analysis
360 (SDTA) SEIKO SSC/5200 SII. Data were collected in air with a flow rate of 2 μl/min, in
361 the range 20–1000 °C, and heating rate of 10 °C/min. Instrumental precision was checked
362 by repeated collections on reference samples, revealing good reproducibility
363 (instrumental theoretical precision of ±0.5 °C); theoretical weight sensitivity is down to
364 0.1 μg. The DTA traces of the reference samples are reported in Viti (2010). The DTA
365 peak deconvolution has been performed following the method described in Viti et al.
366 (2011). Fitting of the DTA peaks was performed using the deconvolution process avail-
367 able in PeakFit4 (PeakFit Version 4.12, SPSS Inc., AISN Software). The temperature
368 range selected for data analysis is 550–800 °C, that is the main dehydroxylation range, in
16
369 the attempt to avoid possible interferences with the 820 °C exothermic signals (Viti
370 2010). The Loess algorithm was used to smooth the data and to remove local
371 perturbations with a ratio of 5%. After that, baseline subtraction was accomplished by a
372 second derivative test procedure.
373
374 Although it is well known that both transmission electron microscopy (TEM) and
375 Raman are powerful diagnostic techniques for the study of asbestos containing materials,
376 the two techniques have not been used in this work where the preference was given to the
377 custom methods specifically developed for the determination of the concentration of
378 asbestos minerals in massive materials.
379
380 RESULTS
381 This section is divided into two parts. The first part summarizes the results of each
382 experimental technique applied to the determination of the chrysotile asbestos content in
383 the natural samples C2 and E1, and the two especially prepared standards STD1 and
384 STD2. The second part describes the results of the quantitative phase analysis of the
385 samples collected at Crètaz and Emarèse sites.
386
387 Optical observations
388 The application of the PCOM technique to the study of our samples was of great help
389 from a qualitative point of view as it easily permitted to identify the nature of fiber
390 bunches thanks to the chromatic dispersion (CD) of the fibers (McCrone 1987; NIOSH
391 1989; EPA 1993; Moss 1994; Bellopede et al. 2009; Cavariani et al. 2010). Figure 2a,b
17
392 shows beautiful fiber aggregates of chrysotile and tremolite. From a quantitative
393 standpoint, because the fiber counting is usually conducted at a low magnification
394 (generally from 40 to 500x), only fibers or fiber aggregates of considerable size can be
395 observed and counted, and it is sometimes difficult to distinguish between an asbestos
396 fiber and crystal of different nature with acicular crystal habit (Fig. 2c). As a matter of
397 fact, fibers with diameter smaller than 0.8-1 μm are not visible with optical microscopy at
398 500x with a resolution power of 0.2 μm (Cavariani et al. 2010).
399
400 SEM analyses
401 The SEM determinations allowed to detect even very small fibers (with a length
402 shorter than 1-2 μm) invisible under optical microscope. We have observed that the
403 sample preparation is crucial and a correct fiber counting is possible only if the stub is not
404 overloaded, that is the case when particle aggregates are formed (see an example in Fig.
405 3a). Small fibers are frequent in the standard samples STD1 and STD2 which were
406 prepared using dry powders. Figure 3b,c reports selected SEM images of the investigated
407 samples showing the presence of fibers with heterogeneous size of both chrysotile and
408 tremolite asbestos fibers. The use of the EDS spot analysis is essential for the correct
409 identification and counting of the chrysotile and tremolite asbestos fibers (Fig. 3c with
410 the relative EDS analysis in the white box). The use of this technique also permits to
411 detect fiber aggregates with distinctive habit and texture and consequently hard to
412 classify otherwise. One such example is portrayed in Figure 3d where an intergrowth of
413 chrysotile-antigorite aggregate is likely observed. Such aggregates may witness low
414 grade regional progressive metamorphism of serpentinite which involves formation of
18
415 antigorite at expenses of chrysotile (Mellini et al. 1987). This paragenesis has already
416 been reported for the Val Malenco serpentinites in the Italian Alps where chrysotile has
417 been identified as antigorite precursor (Mellini et al. 1987; Wicks and O’Hanley 1988).
418 Notwithstanding, the study of the serpentinite texture and its origin is out of the aims of
419 this paper and should be accomplished using TEM.
420
421 FTIR data
422 Figure 4 portrays the selected region between 3400 and 3950 cm-1 of the FTIR spectra
423 of the four investigated samples, where the absorption bands of chrysotile and
424 interference layer silicates occur. The calculated absorbance of the stretching vibration of
425 chrysotile and the resulting chrysotile content (wt%), using the calibration curve with
426 equation y = 0.78269x with y = absorbance (%) and x = chrysotile weight fraction, are
427 reported in Table 1.
428
429 XRPD and the Rietveld method
430 Because the quantitative analysis of chrysotile asbestos using XRPD with the external
431 standard method (Klug and Alexander 1974) requires the determination of the peak
432 intensity or the integrated area of the major diffraction peak of the (001) peak of
433 chrysotile in the unknown sample, the severe peak overlap due to the interference of both
434 antigorite and chlorite, as observed the investigated samples makes this method
435 impracticable. The application of blind single peak fitting procedures to deconvolute each
436 single contribution resulted in unreliable outcomes. For this reason, the use of this
437 method for the determination of the chrysotile asbestos concentration in these matrices
19
438 turns out to be unfeasible. On the other hand, if a more robust fitting procedure with
439 independent observations and (crystallographic) constraints is applied, deconvolution of
440 each contribution seemed viable. As a matter of fact, the application of the Rietveld
441 method guarantees both the use of independent observations (it is a full profile fitting
442 procedure where all the peaks of each phase in the system are modeled) and robust
443 crystallographic constraints (symmetry, unit cell, profile coefficients, background and
444 others). Accuracy of estimates for multiple serpentine minerals can be achieved because
445 all independent reflections belonging to the different serpentine polymorphs in the
446 collected range can be fitted during the refinement procedure. To this aim, areas where
447 overlap of the peaks of chrysotile and antigorite is limited, with distinct observations that
448 help constrain refinement of these phases, exists: see for example the major (13⎯1) peak
449 of antigorite at 2.52 Å which is well separated from chrysotile reflections (⎯2 0 2) and (2
450 0 2), at 2.45 Å and 2.55 Å, respectively. In addition, the fit of a number of peaks of each
451 phase in the systems makes it possible to correct a posteriori any preferred orientation
452 effect, if present, which may bias the major intensities of chrysotile, antigorite and
453 chlorite. The result is a very good fit which allows the calculation of the weight estimates
454 of each phase in the system including chrysotile (see the example of sample E1 in Fig 5).
455 As far as the four analyzed samples, agreement factors of the Rietveld refinements
456 (Larson and Von Dreele 1999) were in the range: Rwp = 9.44-10.33%, Rp = 7.07-8.02%,
457 χ2 = 2.709-5.397. The Rietveld graphical outputs are shown in Figure 6 and Table 2
458 reports the agreement factors and quantitative phase composition.
459
460 DTA analyses
20
461 Figure 7a-d shows the DTA endothermic signals of the four investigated samples and
462 standard pure chrysotile, due to serpentine dehydroxylation processes in the range 550–
463 850 °C and relative result of the fitting procedure. DTA signals were fit using two peaks
464 for chrysotile (dehydroxylation I and II, weak and strong, respectively) and one peak for
465 antigorite (I in Viti et al. 2011). Figure 7e shows the DTA trace and result of the fitting
466 procedure for the pure chrysotile sample. Only fits with a regression coefficient R2 ≥
467 0.990 were considered acceptable. The calculated peak areas were reduced to area ratios
468 (ratio between the calculated area in the unknown and that of the pure chrysotile
469 standard) and plotted in the curve obtained by Viti et al. (2011) for antigorite+chrysotile
470 mixtures with a calculated equation of y = 0.0108x – 0.016. The results of the procedure
471 are reported in Table 3.
472
473 Mineralogical composition and asbestos concentration of the Crètaz and Emarèse
474 samples
475 Agreement factors of the Rietveld refinements (Larson and Von Dreele 1999) were in
2 476 the range Rwp = 6.46-11.32%, Rp = 4.08-8.01%, χ = 3.437–8.639. As an example, the
477 Rietveld graphical output of sample C5 is reported in Figure 8. It should be noted the low
478 peak to background ratio observed in a few samples was due to the fluorescent scattering
479 induced by the large iron content (large magnetite amounts) which excites the copper
480 atoms of the anticathode. Table 4 reports all the quantitative phase compositions of the
481 investigated samples. Because of the low content of magnetite determined in the
482 investigated samples, the microabsorption effects are presumably negligible and no
483 Brindley (1945) correction was applied during the refinement procedure.
21
484
485 DISCUSSION
486 Accuracy of the results from the various experimental techniques
487 The accuracy of the concentration figures of chrysotile asbestos in the abandoned sites
488 of Crètaz and Emarèse was assessed for two selected samples C2 and E1 and the two
489 especially prepared standards STD1 and STD2, with known asbestos concentration, using
490 the experimental techniques permitted by the current Italian regulation for massive
491 materials (XRPD, FTIR, SEM, and PCOM), the Rietveld method and DTA. Table 5
492 summarizes the resulting calculated concentration of chrysotile asbestos in the four
493 samples.
494 Optical microscopy
495 Although the application of the PCOM technique was qualitatively very useful, the
496 figures obtained with this technique are underestimated with respect to the values
497 provided by the application of the other experimental methods (Table 5): an estimate of
498 3.7 wt% was calculated for STD1 sample against an expect value of 5.11 wt% and 1.4
499 wt% for STD2 sample against an expected value of 10.44 wt%. It is possible that limited
500 resolution power did not allow to reckon and quantify small fibers or fiber aggregates
501 predominantly present in the fine powder standards. Analogous conclusions can be found
502 in the literature for the analysis of both airborne and massive sample. As an example, for
503 airborne asbestos fibers in steam tunnels, Dufresne et al. (2002) reported that
504 concentrations found by the TEM/EDS method were higher than those determined by
505 PCOM, especially when chrysotile fibers were present, probably because the TEM/EDS
506 method has a higher resolution than PCOM. Airborne asbestos fibers were present in all
22
507 the steam tunnels and were barely detectable by the PCOM technique only. Our work
508 confirms the limits of the optical methods and it is not surprising that in many countries
509 optical microscopy was adopted only for the visual qualitative identification of asbestos
510 fibers (see for example, EPA 1989; OSHA 1994; HSE 1994).
511 Electron microscopy
512 The chrysotile concentrations obtained from the SEM LAB1 analyses are fairly more
513 accurate than the figures determined from the optical observations. On the other hand, the
514 concentrations provided by SEM LAB2 evidence disagreement (see Table 5). The values
515 provided by LAB1 laboratory are accurate and in agreement with the expected values: an
516 estimate of 4.6 wt% was calculated for STD1 sample against an expect value of 5.11 wt%
517 and 13.1 wt% for STD2 sample against an expected value of 10.44 wt%. Both calculated
518 weights of natural samples C2 and E1 are also in agreement with the values provided by
519 other methods. The values provided by the SEM LAB2 laboratory are largely
520 underestimated as far as the standards are concerned.
521 Due to the higher resolution power and larger field depth, SEM allows to detect small
522 fibers invisible using the optical microscope. The standards were prepared using
523 chrysotile powders with fine chrysotile fibers hard to detect with an optical microscope
524 even at 500x but clearly visible at 2000x with an electron microscope. This may explain
525 the underestimation in the chrysotile concentration determined with the PCOM
526 technique. The difference in the performance of the two microscopic techniques in favor
527 of the SEM method have already been reported by Cavariani et al. (2010). Cavallo and
528 Rimoldi (2013) who studied similar matrices (asbestos serpentinite quarries in
529 Valmalenco, Central Alps, Northern Italy) reported that the small thickness of chrysotile
23
530 fibrils produced during quarrying activities, and the abundance of pseudo-fibrous
531 antigorite cleavage fragments proved the SEM-EDS analytical procedure to be the most
532 suitable.
533 Concerning the results provide by LAB2, Davies et al. (1996) have previously
534 reported that the method for quantitative analysis of asbestos with SEM for both airborne
535 and massive samples presents difficulties, primarily because of the complexity in
536 standardizing the many operating parameters that are controlled. Although some modern
537 SEMs are capable of resolving surface features at the angstrom level, there is still much
538 difficulty in obtaining adequate images of unit fibrils. Such images are necessary to
539 perform quantitative analyses. Moreover, even though routine SEM/EDS analysis allows
540 evaluation of sample morphology and provides semi-quantitative chemical data, it does
541 not allow determination of crystal structure like TEM. Although TEM is a very time-
542 consuming method, seldom providing concentrations that can be considered
543 representative of the whole sample, it is well known that its use makes it possible to
544 resolve many of the ambiguities that lead to incorrect fiber concentration (Beaman and
545 File 1976; Cattaneo et al. 2012).
546 The underestimation of the fiber concentration from LAB2 may be explained by
547 sample powder heterogeneity. We have observed ball of thread like aggregates of
548 chrysotile fibers within the powders (see Fig. 3e) of the samples prepared in LAB2
549 determining areas with a great concentration of asbestos fibers and areas with no fibers at
550 all. A possible explanation is the use of the surfactant sodium lauryl sulfate for the
551 preparation of the suspension in place of dioctyl sodium sulfoccinate. Sodium lauryl
552 sulfate rises the pH of aqueous solution to about 9.5 whereas dioctyl sodium sulfoccinate
24
553 stabilizes pH values of the aqueous solution to about 6.2. In distilled water, the zeta
554 potential of chrysotile displays strongly positive values (about +40 mV) at 6.2 and
555 decreases down to about 20 mV at pH = 9.5 (Light and Wei 1977). The decrease of the
556 zeta potential towards the isoelectric point determines instability of the suspension
557 because the particles with low zeta potential have no force to prevent them coming
558 together and flocculating. Jolicouer et al. (1981) used these principles to retard
559 sedimentation of asbestos fibers in NaCl solutions. Hence, the flocculating chrysotile
560 fibers tend to lump together to form the observed ball of thread like aggregate. On a
561 speculative level, it is not possible to rule out that aggregation also occurs in the dry
562 powders, especially those that have undergone physical long-distance transportation and
563 mechanical shaking. In both situations, fiber segregation may bias the sample
564 homogeneity and the representativeness of the microscopic observation usually
565 performed using a tiny amount of powder.
566 It was already said that the Italian D.M. 09/06/1994 advises the use of SEM when the
567 estimated asbestos concentration is lower than 1wt% giving no indications on the upper
568 concentration limits of application of the method. The concern about asbestos
569 quantitative determination in the case of massive concentration of fibers is related to
570 difficulties for fiber identification and counting due to severe fiber overlap, aggregation
571 and stratification. The results of this study point out that SEM method, if a careful sample
572 preparation, dispersion and representativeness are obtained, can be used even for the
573 quantitative determination of chrysotile asbestos in massive materials with concentrations
574 much larger than 1 wt% with a fairly good accuracy.
575 Infrared spectroscopy
25
576 There are many examples of application of FTIR for the quantitative determination of
577 chrysotile asbestos in bulk materials (see for example, Balducci and Valerio 1986;
578 Massola 1997) although some criticism still exists. According to Davies et al. (1996),
579 FTIR is not a sensitive method for quantifying asbestos in loose aggregates and does not
580 produce useful results when multi-component mixtures are analyzed. Oppositely, a
581 positive report is provided by Foresti et al. (2003) who described the application of the
582 FTIR Linear Calibration Curve method for the determination of low levels (0.01–1 wt%)
583 of free fibers of chrysotile in contaminated clayey, sandy and sandy-organic soils. The
584 detection limit of 0.01 wt% was reached with an enrichment of free fibers of chrysotile in
585 the samples using a standard laboratory elutriator for sedimentation analysis. The
586 linearity of the calibration curves obtained for samples having different soil matrices,
587 indicates that the matrix effects can be accounted for. Recently, De Stefano et al. (2012)
588 compared the accuracy and precision of the Linear Calibration Curve Method and the
589 Method of Addition for the FTIR quantitative determination of asbestos bulk matrices
590 and found that, providing careful samples preparation, both techniques quantify the
591 asbestos content at the level of 1-2 wt% with good precision. The results provided in this
592 study are very important as this is one of the few examples of application of the FTIR
593 quantitative method to complex serpentinite samples.
594 The figures obtained with this method seem rather accurate and in agreement with the
595 values provided by the application of the other experimental methods (Table 5): an
596 estimate of 4.9(2) wt% was calculated for STD1 sample against an expect value of 5.11
597 wt% and 13.8(2) wt% for STD2 sample against an expected value of 10.44 wt%. Both
598 calculated weights of sample STD2 and E1 (32.9 wt% vs. 25.3 wt% from the Rietveld
26
599 method and 27.5 wt% from the SEM analysis) are apparently overestimated. This
600 discrepancy may be possibly due to: (i) interference of antigorite and chlorite present in
601 the investigated samples with a severe overlap of their absorption bands to that of
602 chrysotile; (ii) deviation of the linearity of the curve intensity of the chrysotile absorption
603 band vs. concentration for large chrysotile contents. The breaking of the linearity
604 assumption has been already observed by De Stefano et al. (2012). They observed that in
605 the case of chrysotile, the peak intensity is a quadratic-like function of the concentration
606 in the extended concentration range.
607 X-ray powder diffraction
608 In this work it was found that the XRPD external standard method (Klug and
609 Alexander 1974) is unsuitable for such complex matrices as the observation (e.g., the
610 (001) peak of chrysotile) overlaps to the major peaks of both antigorite (001) and chlorite
611 (002) present in the samples (see sample E1 in Fig. 5), resulting in biased estimates of the
612 integrated area using the single peak fitting deconvolution. The limits of the single peak
613 method for such complex matrices were obvious to Giacomini et al. (2010) who obtained
614 only qualitative estimates of the relative abundance of serpentine polymorphs in samples
615 from the metaophiolites of the the Voltri Massif and Sestri–Voltaggio Zone (Liguria, NW
616 Italy), using the internal standard (20 wt% of corundum) technique and the reference
617 intensity ratio (RIR) method (Snyder and Bish 1989). On the same line, only qualitative
618 determination via XRPD methods were reported by Rigopoulos et al. (2010) for
619 asbestiform minerals in basic and ultrabasic rocks from ophiolite suites of central and
620 northern Greece, Beneduce et al. (2012) for the ophiolites of the Pollino National Park
621 (Calabria-Lucania border, southern Italy), and Cavallo and Rimoldi (2013) for
27
622 serpentinites of the Valmalenco are (Central Alps, Northern Italy). Puledda and Marconi
623 (1990) assessed the validity of the XRPD external standard method with the Ag filter for
624 synthetic mixtures prepared for the determination of chrysotile content in bulk and
625 airborne samples but did not apply it to natural complex samples such as the one
626 investigated here. It is difficult to evaluate the analytical accuracy of the estimates
627 reported by Davis (1990) who used the RIR method to determine the content of asbestos
628 minerals in synthetic multicomponent mixtures prepared samples demonstrate that the
629 lower limit of detection for most asbestos minerals falls in the range 0.5% to 2 wt%.
630 On the contrary, the figures obtained with the Rietveld method seem accurate (Table
631 5): an estimate of 5.8(2) wt% was calculated for STD1 sample against an expect value of
632 5.11 wt% and 11.2(2) wt% for STD2 sample against an expected value of 10.44 wt%.
633 Sample E1 and C2 also contain fairly large amounts of chrysotile asbestos. The results
634 reported here witness that the Rietveld method can be successfully applied for the
635 determination of chrysotile asbestos in these complex multicomponent samples
636 characterized by the presence of two serpentine polymorphs (antigorite and chrysotile)
637 and chlorite. It is not possible to generalize the outcome of this study and assert that the
638 method is accurate even for more complex samples which contain all the serpentine
639 polymorphs including lizardite, although this mineral association is not rare in green
640 stones. Another critical aspect regards the range of applicability of the Rietveld method in
641 terms of chrysotile content. It is well known that chrysotile possesses a cylindrical lattice
642 with layers curled concentrically or spirally, usually around the x axis (clinochrysotile
643 and orthochrysotile) and seldom around the y axis (parachrysotile), to form a tubular
644 structure (roll) of about 22±27 nm in diameter (Whittaker 1957; Yada 1971). Moreover,
28
645 the cylindrical lattice also displays extensive structural defectivity (e.g., random shift
646 component along the fiber axis to yield cone lattices). Such structure complexity results
647 in anisotropic broadening effects of the peak profiles in the X-ray powder patterns which
648 can only be properly fit by recursive models such as the one implemented in DIFFaX+
649 (Leoni et al. 2004) and not simply by Rietveld-based deterministic codes. For this reason,
650 Wilson et al. (2006) developed and alternative approach to determine the chrysotile
651 content in mine tailings at Clinton Creek (Yukon Territory) and Cassiar (British
652 Columbia) using structure-less pattern fitting and with the addition of a known quantity
653 of a well-crystallized material, by considering the serpentine minerals as amorphous
654 phases, and tested the accuracy of the method using synthetic serpentine-rich mine
655 tailings of known composition. Notwithstanding, marked peak profile broadening effects
656 (especially the asymmetric band in the 19-26 °2θ range due to the destructive interference
657 of non-basal diffraction peaks) are obvious when chrysotile fraction is high. Usually,
658 when chrysotile is present in medium-low concentration, the powder pattern generally
659 displays only the major basal (00l) peaks of chrysotile and minor non-basal peaks. Thus,
660 peak profile broadening effects are less obvious and may be modeled by empirical
661 Lorentzian profile anisotropic broadening functions such as the one implemented in
662 GSAS (Larson and Von Dreele 1999): γ = Y + Yecosφ with γ = Lorentzian contribution to
663 the pseudo-Voigt profile function; Y = isotropic strain broadening term; Ye = anisotropic
664 strain broadening term; φ = angle which defines the direction of the broadening axis
665 within the crystallographic setting. Wilson et al. (2006) reported chrysotile estimates as
666 high as 90.8 wt%. For those samples resembling monophasic systems, the accuracy of the
667 Rietveld method is expected to be low. On the other hand, for the systems investigated
29
668 here with chrysotile contents below 40 wt%, the accuracy of the Rietveld method is still
669 expected to hold. As a matter of fact, the refinement of the anisotropic strain broadening
670 term of the Lorentzian part of the pseudo-Voigt function was attempted for chrysotile for
671 samples C2 and E1, setting the broadening direction along the b axis and the stacking
672 fault model but did not significantly changed the final calculated weights.
673 Thermal analysis
674 The quantitative concentration of chrysotile in the standard samples using the DTA
675 method, reported in Table 5, are rather accurate: 7.6 wt% was calculated for STD1
676 sample against an expect value of 5.11 wt% and 12.2 wt% for STD2 sample against an
677 expected value of 10.44 wt%. Although the successful application of the method has
678 already been reported in the literature for simple systems such as pharmaceutical grade
679 talc where the minimum level of detection was 1% by weight of chrysotile asbestos
680 (Shelz 1974) and in cosmetic grade talc/body products where the minimum level of
681 detection was 0.5% by weight of chrysotile asbestos (Luckewicz 1975), only the recent
682 contribution of Viti et al. (2011) reported accurate estimates for complex natural systems
683 such as serpentinites. Hence, the results of this work are a further confirmation of the
684 reliability of this method. The apparently systematic overestimation should be due to the
685 presence of chlorite in the system. As a matter of fact, Viti et al. (2011) reported that the
686 possible occurrence of other minerals does not hamper the attainment of reliable
687 qualitative determinations, with the only exception of chlorite whose typical DTA curve
688 is characterized by an endothermic peak at 600–650 °C (Smykatz-Kloss 1974), and
689 suggested that, prior to DTA data collection, qualitative XRPD should be performed to
690 ascertain the presence and the amount of chlorite. The presence of chlorite was assessed
30
691 in the first instance via XRPD suggesting the addition of a further peak due to chlorite
692 during the process of fit deconvolution. Notwithstanding, it is possible that a systematic
693 correlation occurred for the calculation of the integrated areas and that the resulting
694 integrated area of chrysotile was underestimated in favor of that of chlorite.
695
696 Crètaz and Emarèse samples
697 According to the Italian law D.Lgs 12/03/2010 nr. 205 All. D, if the asbestos
698 concentration is higher than 0.1 wt%, a site should be classified as “contaminated by
699 asbestos”. Hence, both Crètaz and Emarèse sites should be reclaimed as all the collected
700 samples consistently show asbestos concentration much higher than the imposed limit
701 (Tables 4 and 5).
702 Figure 9a,b reports maps of concentration of chrysotile and tremolite obtained from
703 the Rietveld analysis for the site of Crètaz (Tables 4 and 5). The maps were obtained by
704 geostatistical analysis of the raw data in GIS environment using the software ArcGIS
705 10.1 and the ArcGIS Geostatistical Analyst application. The Inverse Distance Weighting
706 (IDW) model (Shepard 1968) was used for the deterministic interpolation of the data
707 points. For the chrysotile concentration maps, the following parameters were used: first
708 order model (power = 1), maximum neighbors = 10, minimum neighbors = 5, total
709 calculation area = circular with radius of 138 m. For the tremolite concentration maps, the
710 following parameters were used: second order model (power = 2), maximum neighbors =
711 10, minimum neighbors = 5, total calculation area = circular with radius of 200 m.
712 The chrysotile content varies from 4.4 to 22.8 wt%. In all the samples, chrysotile is
713 associated to antigorite (6.3-50.1 wt%), chlorite (clinochlore 4.1-15.2 wt%), muscovite
31
714 (2.2-15.8 wt%), quartz (1.9-29.1 wt%), and calcite (4.2-27.5 wt%). The mineral
715 assemblage reflects the overall composition of the mine tailings, a mixing of three
716 different rock types all belonging to the Piedmont Zone unit: the antigorite-serpentinites
717 with magnetite formation, the calc-schists s.l. and local quartzites formation, and the
718 exotic continental Permian-mesozoic succession with schists, quartzites, marbles, and
719 limestones. The distinctive minerals of the latter formation are quartz, mica, carbonates
720 calcite and dolomite, plagioclase, K-feldspar and tremolite (1.8-10.3 wt%). Apparently
721 no trends of the chrysotile and tremolite concentration with (horizontal or vertical) space
722 are observed. This random distribution is due to the history of the various deposits
723 emplaced and mixed with other source of rock dumps during the activity of the mine and
724 not to physical factors such as fiber leaching, transport, and re-deposition. The area with
725 the highest surface concentration of asbestos minerals (see Figure 9) should be
726 considered as priority of intervention in the reclamation plan.
727 Figure 9c, not elaborated with the geostatistical analysis due to the limited number of
728 raw data points, reports the concentration of chrysotile in the site of Emarèse as obtained
729 from the Rietveld analysis (see also Tables 4 and 5). The chrysotile content varies from
730 3.3 to 39.5 wt% with the concentration linearly decreasing with the distance from the
731 main mining sites. Chrysotile is invariably accompanied by antigorite (3.3-44.7 wt%) and
732 chlorite (clinochlore 12.4-32.6 wt%). On the basis of the mineral assemblage, two groups
733 of rock samples are observed: (i) samples E1-E5 belongs to the antigorite-serpentinites
734 with magnetite formation of the Piedmont Zone. Besides the typical serpentinite phases
735 antigorite, chrysotile and chlorite, magnetite, talc and brucite are found together with the
736 minerals of the original ultramaphic rock forsterite and enstatite. Hematite, calcite and
32
737 dolomite are likely secondary phases formed after surface alteration in contact with
738 water; (ii) samples E6-E7 are collected from the surface debris and soil composed of a
739 mixing of the former antigorite-serpentinites with magnetite formation and the calc-
740 schists s.l. and local quartzites formation both belonging to the Piedmont Zone unit. The
741 distinctive minerals of the calc-schists s.l.- quartzite formation are quartz, mica, K-
742 feldspar, plagioclase, and tremolite (5.9-12.4 wt%).
743 The area with the highest surface concentration of chrysotile (see Figure 9c) should
744 also be classified as prioritary in the reclamation plan.
745
746 CONCLUDING REMARKS
747 A map of the concentration of asbestos minerals chrysotile and tremolite in
748 contaminated mine tailings of the Valle d’Aosta region (Northern Italy) was drawn. The
749 results of this study are strategic as the knowledge of the spatial distribution of the
750 asbestos concentration allows to plan reclamation agenda of the sites. The area with the
751 highest surface concentration of both chrysotile and tremolite were identified and
752 classified as priority in the reclamation plan. The results of our study have general
753 implications as the protocol of sampling, data analysis and mapping of the asbestos
754 concentration can be used in any region of the globe where such deposits are found.
755 Besides that, another general implication of our work regards the assessment of the
756 accuracy of the results obtained using state of the art experimental techniques for the
757 determination of the concentration of asbestos in massive materials. Although we focused
758 on the methods recommended by the Italian laws, the Rietveld method and DTA (Viti et
759 al. 2011) were also included so that a comparison among the various methods used
33
760 worldwide was possible. Accurate estimates were obtained using the SEM, FTIR, XRPD
761 Rietveld and DTA methods whereas poor figures were obtained with the optical
762 microscopy. The single peak fitting XRPD method revealed to be unsuitable because of
763 the interference effects of antigorite and chlorite. It should be remarked that the Rietveld
764 method has been successfully applied for the determination of chrysotile concentrations
765 as high as about 40 wt% in these complex serpentinite multi-component samples
766 characterized by the presence of two serpentine polymorphs (antigorite and chrysotile)
767 and chlorite. Further investigations are required to test the accuracy of the method for
768 even more difficult systems which contain all the three serpentine polymorphs, including
769 lizardite. With higher concentrations (>50 wt%?), it is recommended to apply other
770 methods such as the one proposed by Wilson et al. (2006) who determined the chrysotile
771 content in asbestos mine tailings using structure-less pattern fitting and with the addition
772 of a known quantity of a standard material so to be able to consider the serpentine
773 minerals as amorphous phases.
774
775 Acknowledgments
776 Thanks are due to the technical staff of the C.I.G.S. Laboratory (The University of
777 Modena and R.E., Modena Italy). Thanks to the staff of the CGT - Centro di Geo
778 Tecnologie) University of Siena, Italy for help with the FTIR analyses. Thanks to A.
779 Facchinetti and A. Zanella (ARPA Valle d'Aosta, Sezione Analisi strutturali e amianto)
780 for their help with the SEM analyses. This manuscript benefited greatly from the accurate
781 revisions of Dr. S.A. Wilson and suggestions of the Associate Editor Prof. M. Gunter.
782
34
783 REFERENCES CITED 784 785 Bailey, K. F. (1988) Hydrous Phyllosilicates (exclusives of micas) Vol. 19. Series editor 786 P. H. Ribbe. Mineralogical Society of America, pp., 518. 787 788 Balducci, D. and Valerio, F. (1986) Qualitative and Quantitative Evaluation of Chrysotile 789 and Crocidolite Fibers with IR-Spectroscopy: Application to Asbestos-Cement 790 Products. International Journal of Environmental Analytical Chemistry 27(4), 315- 791 323. 792 793 Beaman, D.R. and File, D.M. (1976) Quantitative determination of asbestos fiber 794 concentrations. Analytical Chemistry, 48, 101–110. 795 796 Bellopede, R., Clerici, C., Marini, P. and Zanetti, G. (2009) Rocks with Asbestos: Risk 797 Evaluation by Means of an Abrasion Test. American Journal of Environmental 798 Sciences, 5(4), 500-506. 799 800 Beneduce, P., Di Leo, P., Ivo Giano, S. and Schiattarella, M. (2012) Quantitative 801 geomorphic analysis of asbestos dispersion and pollution from natural sources: the 802 case-history of the Pollino national park, southern Italy. Geografia fisica e Dinamica 803 del Quaternario (in Italian), 35, 3-12. 804 805 Birle, J. D., Gibbs, G. V., Moore, P. B. and Smith, J. V. (1968) Crystal structures of 806 natural olivines. American Mineralogist, 53, 807-824. 807 808 Brindley, G. W. (1945) The effect of particle size and microabsorption on diffracted 809 powder line intensities. Philosophical Magazine, 36, 347-369. 810 811 CARB Method 435, (1991) Determination of asbestos content of serpentinite aggregate. 812 State of Air Resources Board,California. 813 http://www.arb.ca.gov/testmeth/vol3/M_435.pdf
35
814 815 Cattaneo, A., Somigliana, A., Gemmi, M., Bernabeo, F., Savoca, D., Domenico, M. 816 Cavallo, and P.A., Bertazzi (2012) Airborne Concentrations of Chrysotile Asbestos in 817 Serpentine Quarries and Stone Processing Facilities in Valmalenco, Italy. Annals of 818 Occupational Hygiene, 56(6), 671–683. 819 820 Cavallo, A. and Rimoldi, B. (2013) Chrysotile asbestos in serpentinite quarries: a case 821 study in Valmalenco, Central Alps, Northern Italy. Environmental Science: Processes 822 & Impacts, in press. 823 824 Cavariani, F., Marconi, A. and Sala, O. (2010) Asbestos: sampling, analytical techniques 825 and limit values. Italian Journal of Occupational and Environmental Hygiene, 1(1), 826 18-28. 827 828 Cavinato, A. (1964) I Giacimenti minerari. Torino : UTET, stampa 1964. - XLVII, pp. 829 686 (in Italian). 830 831 Collegium Ramazzini (2010), ‘Asbestos is still with us: repeat call for universal ban’, 832 Occupational Medicine, 60, 584–588. 833 834 Dal Piaz, G.V. (1971) Alcune considerazioni sulla genesi delle ofioliti piemontesi e dei 835 giacimenti ad esse associati. Bollettino Associazione Mineralogica Subalpina, Torino, 836 8: 365-388 (in Italian). 837 838 Davies, L.S.T., Wetherill, G.Z., McIntosh, C., McGonagle, C., and Addison, J. (1996) 839 Development and validation of an analytical method to determine the amount of 840 asbestos in soils and loose aggregates. HSE Contract research Report No. 83/1996, pp. 841 69. 842 843 Davis, B.L. (1990) Quantitative Analysis of Asbestos Minerals by the Reference Intensity 844 X-Ray Diffraction Procedure. American Industrial Hygiene Association Journal,
36
845 51(6), 297-303. 846 847 De Stefano, L., Cioffi, R. and Colangelo, F. (2012) Comparison between Two FTIR 848 Spectroscopy Analytical Procedures for Micrograms Determination of Asbestos 849 Species in Bulk Materials. American Journal of Analytical Chemistry, 3, 1-5.
850 851 Di Colbertaldo, D., Di Furia, E. and Rossi, F. (1967) Il giacimento a magnetite di Cogne 852 in Val d’Aosta. Rendiconti Istituto Lombardo, 101, 361-394 (in Italian). 853 854 Diella, V., Ferrario, A. and Rossetti, P. (1994) The magnetite ore deposits of the southern 855 Aosta Valley: chromitite transformed during an alpine metamorphic event. Ofioliti, 856 19, 247-256. 857 858 Dilek, Y. and Newcomb, S. (Eds.) (2003) Ophiolite concept and the evolution of 859 geological thought (No. 373). Geological society of America, pp. 505. 860 861 Dollase, W.A. (1986) Correction of intensities for preferred orientation in powder 862 diffractometry: application of the March model. Journal of Applied Crystallography, 863 19, 267-272. 864 865 Downs, R.T., Hazen, R.M. and Finger, L.W. (1994) The high-pressure crystal chemistry 866 of low albite and the origin of the pressure dependency of Al-Si ordering. American 867 Mineralogist, 79, 1042-1052. 868 869 Dufresne, A., Anthes, M., Santos, B. and Infante-Rivard, C. (2002) Airborne Asbestos 870 Fibres in Steam Tunnels: A Qualitative Analysis by PLM and a Quantitative Analysis 871 by PCOM and TEM/EDS. Annals of Occupational Hygiene, 46(1), 125–127. 872 873 Elter, G. (1960) La zona pennidica dell’alta e media Valle d’Aosta e le sue unità 874 limitrofe. Memorie Istituto Geologia e Mineralogia Università di Padova (Italy), 22,
37
875 113 (in Italian). 876 877 Environmental Protection Agency (EPA) (1989) Code of Federal Regulations, Pt. 763, 878 Appendix A- Interim method of the determination of asbestos in bulk insulation 879 samples, July 1989. 880 881 Environmental Protection Agency (EPA) (1993) Test Method - Method for the 882 determination of asbestos in bulk building materials. EPA/600/R-93/116, July 1993. 883 884 Falini, G., Foresti, E., Gazzano, M., Gualtieri, A.F., Leoni, M., Lesci, I.G. and Roveri, N. 885 (2004) Tubular-shaped stoichiometric chrysotile nanocrystals. Chemistry European 886 Journal, 10, 3043–3049. 887 888 Farmer, V.C. (1974) The Infrared Spectra of Minerals. The Mineralogical Society, 889 London, U.K, pp. 539. 890 891 Foresti, E., Gazzano, M., Gualtieri, A.F., Lesci, I.G., Lunelli, B., Pecchini, G., Renna, E. 892 and Roveri, N. (2003) Determination of low levels of free fibres of chrysotile in 893 contaminated soils by X-ray diffraction and FTIR spectroscopy. Analytical and 894 Bioanalytical Chemistry, 376, 653–658. 895 896 Giacomini, F., Boerio, V., Polattini, S., Tiepolo, M., Tribuzio, R. and Zanetti, A. (2010) 897 Evaluating asbestos fibre concentration in metaophiolites: A case study from the 898 Voltri Massif and Sestri-Voltaggio Zone (Liguria, NW Italy). Environmental Earth 899 Sciences, 61, 1621–1639. 900 901 Gualtieri, A.F. (2000) Accuracy of XRPD QPA using the combined Rietveld–RIR 902 method. Journal of Applied Crystallography, 33, 267–278. 903
38
904 Gualtieri, A. F. (2012) : Mineral fibre-based building materials and their health hazards. 905 Chapter 8 nel volume “Toxicity of Building Materials” (pp. 486), Edited by F. 906 Pacheco-Torgal, S. Jalali and A. Fucic, Woodhead Publishing, 166-195. 907 908 Haavik, C., Stølen, S., Fjellvåg, H., Hanfland, M. and Häusermann, D. (2000) Equation 909 of state of magnetite and its high-pressure modification: Thermodynamics of the Fe-O 910 system at high pressure. American Mineralogist, 85(3-4), 514-523. 911 912 Hodgson, J. T. and Darnton, A. (2000) The quantitative risks of mesothelioma and lung 913 cancer in relation to asbestos exposure. Annals of Occupational Hygiene, 44 (8), 565- 914 601. 915 916 IARC (International Agency for Research on Cancer) (1977) IARC Monographs on the 917 Evaluation of the Carcinogenic Risk of Chemicals to Man: Asbestos. IARC 918 Monographs on the Evaluation of Carcinogenic Risk of Chemicals to Man, 14, 1-106. 919 920 ISO (International Standard Organization) (2002) Ambient air determination of 921 numerical concentration of inorganic fibrous particles scanning electron microscopy 922 method 14966. Geneva, Switzerland. 923 924 Jolicoeur, C. and Duchesne, D. (1981) Infrared and thermogravimetric studies of the 925 thermal degradation of chrysotile asbestos fibres: evidences for matrix effects. 926 Canadian Journal of Chemistry, 59, 1521–1526. 927 928 Klug, H. P. and Alexander, L. E. (1974) X-ray Diffraction. Addision-Wilson Publishing 929 Company Inc, USA, 132 pp. 966. 930 931 Larson, A.C. and Von Dreele, R.B. (1999) Generalized Structure Analysis System. Los 932 Alamos National Lab, New Mexico, LAUR 86-748. 933
39
934 Leoni, M., Gualtieri, A.F., and Roveri, N. (2004) Simultaneous refinement of structure 935 and microstructure of layered materials. Journal of Applied Crystallography, 37, 166– 936 173. 937 938 Le Saoût, G, Kocaba, V. and Scrivener, K. (2011) Application of the Rietveld method to 939 the analysis of anhydrous cement. Cement and Concrete Research, 41, 133-141. 940 941 Light, W. G. and Wei, E. T. (1977) Surface charge and hemolytic activity of asbestos. 942 Environmental Research, 13(1), 135-145. 943 944 Luckewicz, W. (1975) Differential thermal analysis of Chrysotile asbestos in pure talc 945 and talc containing other minerals. Journal of the Society of Cosmetics Chemistry, 26, 946 431-437. 947 948 Massola, A. (1997) Campionamento e analisi di materiali contenenti amianto. L’amianto: 949 dall’ambiente di lavoro all’ambiente di vita. Nuovi indicatori per futuri effetti a cura di 950 Minoia C., Scansetti G., Piolatto G., Massola A. Laboratorio di Igiene Ambientale e 951 Tossicologia Industriale, Fondazione “S. Maugeri”. Clinica del Lavoro e della 952 Riabilitazione, (I.R.C.C.S.). Centro Medico di Pavia. I Documenti. 12 (in Italian). 953 954 McCrone, W.C. (1987) Asbestos identification: Chicago, Illinois, McCrone Research 955 Institute, G&C Printers Inc., pp. 199. 956 957 Mellini, M., Trommsdorff, V. and Compagnoni, R. (1987) Antigorite polysomatism: 958 Behaviour during progressive metamorphism. Contributions to Mineralogy and 959 Petrology, 97, 147–155. 960 961 Meyer, D. R. (1980) Nutritional problems associated with the establishment of vegetation 962 on tailings from an asbestos mine. Environmental Pollution (Series A), 23, 287 – 298 963 964 Moss, R.D. (1994) Evaluation of asbestos quantitation methods. Microscope, 42, 7-14.
40
965 966 Nicholson W. J. (1986) Airborne asbestos health assessment update. NTIS, 967 SPRINGFIELD, VA(USA). 968 969 NIOSH (National Institute of Occupational Safety and Health) (1989) Asbestos (bulk), 970 Method 9002: Microscopy, Stereo and Polarized Light with Dispersion Staining. 971 NIOSH Manual of Analytical Methods, 5/15/89. 972 973 OSHA (Occupational Health and Safety Administration) (1994) Occupational Exposure 974 to Asbestos. Federal Register 59, 40964, August 10, 1994. 975
976 Perdikatsis, B. and Burzlaff, H.(1981) Strukturverfeinerung am talk Mg3[(OH)2Si4O10]. 977 Zeitschrift für Kristallographie, 156(3-4), 177-186. 978 979 Prince, E., Donnay, G., Martin, R.F. (1973) Neutron diffraction refinement of an ordered 980 orthoclase structure. American Mineralogist, 58, 500-507. 981 982 Puledda, S. and Marconi, A. (1990) Quantitative X-ray diffraction analysis of asbestos by 983 the silver membrane filter method. Application to chrysotile. Am. Ind. Hyg. Assoc. J., 984 51, 107-114. 985 986 Ribbe, P.H. (1979) The structure of a strained intermediate microcline in cryptoperthitic 987 association with twinned plagioclase. American Mineralogist, 64, 402-408. 988 989 Rietveld, H.M. (1969) A profile refinement method for nuclear and magnetic structures. J 990 Applied Crystallography, 2, 65–71. 991 992 Rigopoulos, I., Tsikouras, B., Pomonis, P., Karipi, S. and Hatzipanagiotou, K. (2010) 993 Quantitative analysis of asbestos fibres in ophiolitic rocks used as aggregates and 994 hazard risk assessment for human health. Bulletin of the Geological Society of Greece, 995 Proceedings of the 12th International Congress, Patras, May, 2010
41
996 997 Shelz, J. P. (1974) The detection of chrysotile asbestos at low- levels of talc by 998 differential thermal analysis. Thermochimica Acta, 197-204. 999 1000 Shepard, D. (1968) A two-dimensional interpolation function for irregularly spaced data. 1001 Proceedings of the 1968 ACM National Conference, 517–524. 1002 1003 Skinner, H.C.W., Ross, M. and Frondel, C. (1988) Asbestos and other fibrous materials - 1004 mineralogy, crystal chemistry and health effects: New York, Oxford, Oxford 1005 University Press, pp. 204. 1006 1007 Smykatz-Kloss, W. (1974) Differential Thermal Analysis. Application and results in 1008 mineralogy, Springer-Verlag, Berlin, pp.185. 1009 1010 Snyder, R.L. and Bish, D.L. (1989) Quantitative Analysis, in Bish, D.L. and Post, J.E., 1011 eds., Modern Powder Diffraction, Reviews in Mineralogy, v. 20, Washington, D.C., 1012 Mineralogical Society of America, 101-144. 1013 1014 Toby, B.H. (2001) EXPGUI, a graphical user interface for GSAS. Journal of Applied 1015 Crystallography, 34, 210–213. 1016 1017 Uehara, S. (1998) TEM and XRD study of antigorite superstructures. The Canadian 1018 Mineralogist, 36, 1595-1605. 1019 1020 Ungaretti, L. and Oberti, R. (2000) Leverage analysis and structure refinement of 1021 minerals. American Mineralogist, 85(3-4), 532-542. 1022 1023 Viti, C. (2010) Serpentine minerals discrimination by thermal analysis. American 1024 Mineralogist, 95, 631–638. 1025
42
1026 Viti, C. and Mellini M. (1996) Vein antigorites from Elba Island, Italy. European Journal 1027 of Mineralogy, 8, 423–434. 1028 1029 Viti, C., Giacobbe, C. and Gualtieri, A.F. (2011) Quantitative determination of chrysotile 1030 in massive serpentinites using DTA: Implications for asbestos determinations. 1031 American Mineralogist, 96, 1003–1011. 1032 1033 1034 Wenk, H.R., Joswig ,W., Tagai, T., Korekawa, M. and Smith, B.K. (1980) The average 1035 structure of An 62-66 labradorite. American Mineralogist, 65, 81-95. 1036 1037 Wicks, F. J. and O'Hanley, D. S. (1988) : Serpentine minerals; structures and petrology 1038 Reviews in Mineralogy and Geochemistry, 19, 91-167. 1039 1040 Winburn, R.S., Grier, D.G., McCarthy, G.J. and Peterson, R.B. (2000) Rietveld 1041 quantitative X-ray diffraction analysis of NIST fly ash standard reference materials. 1042 Powder Diffraction, 15, 163-172. 1043 1044 Whittaker, E.J.W. (1957) The structure of Chrysotile, V. diffuse reflexions and fibre 1045 texture Acta Crystallographica, 10, 149-156. 1046 1047 WHO (World Health Organization) (1997) Determination of airborne fiber number 1048 concentrations; a recommended method, by phase contrast microscopy (membrane 1049 filter method). Geneva, Switzerland. 1050 1051 Wilson, S. A., Raudsepp, M. and Dipple, G. M. (2006) Verifying and quantifying carbon 1052 fixation in minerals from serpentine-rich mine tailings using the Rietveld method with 1053 X-ray powder diffraction data. American Mineralogist, 91, 1331–1341. 1054 1055 Yada K. (1971) Study of microstructure of chrysotile asbestos by high-resolution electron 1056 microscopy, Acta Crystallographica A, 27(6), 659-664.
43
1057 1058 Zanazzi, P.F., Comodi, P., Nazzareni, S., and Andreozzi, G.B. (2009) Thermal behaviour 1059 of chlorite: an in situ single-crystal and powder diffraction study. European Journal of 1060 Mineralogy, 21(3), 581-589. 1061
44
1062 Figure Captions
1063 FIGURE 1. Sketch map of the geology of Valle d’Aosta (1:1.000.000 scale; Legend:
1064 Austroalpine System (1 = Diorite–Kinzigite Zone, 2 = Gneiss Complex, 3 = Eclogitic
1065 Micaschist Complex); Piedmont Zone with Calc-schists and Greenstones (4 = Oceanic
1066 Metasedimentary overlay, 5 = Metabasites); Pennidic System (6 = Upper Pennidic Zone,
1067 7 = Middle Pennidic Zone, 8 = Outer Pennidic Zone; 9 = Ultra-elvetic System; 10 =
1068 Elvetic System) and location of the surface and core collection spots for the two
1069 investigated sites: (a) Crètaz with C1-C8 = surface samples, depth 0.25-0.3 m; C9v1 =
1070 drill 1, depth 0.7 m; C10v1 = drill 1, depth 7.5 m; C11v1 = drill 1, depth 15 m; C12v1 =
1071 drill 1, depth 19.2 m; C13v2 = drill 2, depth 0.5 m; C14v2 = drill 2, depth 11 m; C15v2 =
1072 drill 2, depth 22.4 m. (b) Emarèse with E1-E8 = surface samples.
1073 FIGURE 2. Selected PCOM images of the investigated samples (see text for details).
1074 Legend: (a) chrysotile at 10x in liquid with refractive index of 1.550; (b) tremolite fibers
1075 at 10x in liquid with refractive index of 1.615; (c) non-asbestos acicular-like crystal:
1076 diopside at 10x in liquid with refractive index of 1.670.
1077 FIGURE 3. Selected SEM images of the investigated samples (see text for details).
1078 Legend: (a) particle and fiber aggregates difficult to detect and count; (b) chrysotile fibers
1079 of heterogeneous size; (c) tremolite fibers with the relative EDS point analysis which
1080 permits the discrimination from other fibrous phases; (d) example of intergrowth of
1081 mixed chrysotile-antigorite aggregate; (e) ball of thread like aggregates of chrysotile
1082 fibers within the powders of the samples prepared in LAB2 (see the text for details).
-1 1083 FIGURE 4. The FTIR spectra region between 3400 and 3950 cm of the four
1084 investigated samples where the absorption bands of chrysotile and interference layer
45
1085 silicates occur. (a) STD1; (b) STD2; (c) C2; (d) E1.
1086 FIGURE 5. Selected low angle region of sample E1 showing severe overlap of the
1087 peaks of chrysotile, antigorite and clinochlore and the result of the peak fitting using the
1088 Rietveld method.
1089 FIGURE 6. The Rietveld graphical outputs of the samples selected for the accuracy
1090 verification. Crosses represent the observed pattern, the thin line represents the calculated
1091 pattern and the grey bottom line is the difference line. Markers of all the peaks of the
1092 each crystalline phase included in the refinement procedure are also shown in different
1093 lines. Legend: (a) STD1, lines of peak markers from the bottom: antigorite, chrysotile,
1094 tremolite, talc, clinochlore, calcite, dolomite, magnesite, albite. (b) STD2, lines of peak
1095 markers from the bottom: antigorite, chrysotile, tremolite, talc, clinochlore, calcite,
1096 dolomite, magnesite, albite; (c) C2, lines of peak markers from the bottom: antigorite,
1097 chrysotile, tremolite, muscovite, clinochlore, calcite, quartz, K-feldspar, albite; (d) E1,
1098 lines of peak markers from the bottom: antigorite, chrysotile, clinochlore, calcite,
1099 magnetite, forsterite, hematite, enstatite, brucite.
1100 FIGURE 7. DTA traces of the four investigated samples and standard pure chrysotile,
1101 in the range 550–850 °C (see the text for details) with the result of the deconvolution
1102 fitting procedure. Legend: (a) STD1; (b) STD2; (c) C2; (d) E1; (e) pure chrysotile.
1103 FIGURE 8. The Rietveld graphical output of sample C5. Crosses represent the observed
1104 pattern, the thin line represents the calculated pattern and the grey bottom line is the
1105 difference line. Markers of all the peaks of the each crystalline phase included in the
1106 refinement procedure are shown. Legend from the bottom line: antigorite, chrysotile,
1107 tremolite, muscovite, clinochlore, calcite, quartz, K-feldspar, dolomite.
46
1108 FIGURE 9. Maps of concentration of chrysotile (a) and tremolite (b) obtained from the
1109 Rietveld analysis for the site of Crètaz obtained by geostatistical analysis of the raw data
1110 in GIS environment using the software ArcGIS 10.1 and the ArcGIS Geostatistical
1111 Analyst application. The raw concentrations of chrysotile for the site of Emarèse are
1112 plotted in (c).
1113
47
1114 Tables
1115 TABLE 1. The calculated absorbance (%)of the stretching vibration of chrysotile and the
1116 resulting chrysotile content (wt%), using the calibration curve with equation y =
1117 0.78269x with y = absorbance (%) and x = chrysotile weight fraction.
chrysotile content chrysotile content Sample absorbance (%) (wt%) error STD1 3.856 4.9 0.03 STD2 10.801 13.8 0.02 C2 4.931 6.3 0.02 E1 25.75 32.9 0.02 1118
1119
1120
1121
1122
1123
1124
1125
48
1126 TABLE 2. The results of the Rietveld quantitative phase composition. The definition of
1127 the agreement factors is reported in Larson and Von Dreele (1999).
STD1 STD2 C2 E1
Rwp (%) 10.3 9.73 10.33 9.44
Rp (%) 8.02 7.51 7.57 7.07 χ2 5.40 2.71 5.40 6.32 antigorite 6.9(3) 14.9(3) 3.5(3) 44.7(6) brucite - - - 0.3(1) calcite 39.3(2) 36.7(3) 8.0(3) 2.6(2) chrysotile 5.8(2) 11.2(3) 6.3(4) 25.1(9) clinochlore 4.6(2) 5.9(3) 7.5(3) 15.8(4) dolomite 11.9(2) 11.7(2) - - hematite - - 1.3(2) enstatite - - 1.2(3) forsterite - - 4.0(6) K-feldspar - - 4.2(3) - magnesite 4.3(2) 4.6(2) - - magnetite - - - 5.0(2) muscovite/Illite - - 10.9(5) - plagioclase 18.3(2) 10.8(3) 22.4(3) - quartz - - 32.6(2) - talc 3.3(5) 3.2(3) - - tremolite 5.6(3) 1.0(3) 4.6(3) - TOT 100 100 100 100 1128
1129
49
1130 TABLE 3. The results of the chrysotile determination using the DTA method (see the text
1131 for details).
Calculated Sum of the Sample Error r2 wt% Area ratio Error peak area chrysotile Pure chrysotile 1358 39 0.9909 100 1 n.d. standard STD1 89 16 0.9927 7.6 0.06 0.03
STD2 155 8 0.9921 12.2 0.11 0.02
C2 108 3 0.9952 8.9 0.08 0.07
E1 300 26 0.9955 22.2 0.22 0.07 1132 1133 1134
1135
50
1136 TABLE 4. The quantitative phase compositions of the Crètaz and Emarèse samples. The definition of the agreement factors is reported
1137 in Larson and Von Dreele (1999).
C1 C2 C3 C4 C5 C6 C7 C8 C9V1 C10V1 C11V1 C12V1
Rwp (%) 7.68 10.33 6.46 8.8 5.64 11.32 7.12 6.98 7.8 7.86 3.29 10.48
Rp (%) 5.58 7.57 4.81 6.22 4.08 8.01 5.51 5.42 5.94 6.02 5.55 7.93 χ2 5.55 5.40 3.44 6.60 3.87 8.64 2.55 2.96 3.33 3.35 2.64 5.61 antigorite 32.4(8) 3.5(3) 29.5(6) 47.1(6) 39.3(7) 50.1(6) 30.9(5) 23.6(4) 40.4(4) 19.5(7) 14.3(4) 52.4(5) calcite 20.5(3) 8.0(3) 4.2(2) 9.0(3) 11.5(3) 8.1(3) 12.0(2) 8.0(2) 14.7(2) 27.5(2) 16.6(2) 5.0 (2) chrysotile 6.2(9) 6.3(4) 4.4(6) 11.4(8) 9.7(8) 11.2(7) 15.8(4) 11.6(4) 11.8(4) 12.1(5) 6.0(3) 22.8(5) clinochlore 13.4(5) 7.5(3) 10.0(3) 12.0(6) 6.1(5) 11.4(5) 15.2(5) 7.0(3) 10.1(5) 9.4(6) 9.0(3) 7.9(7) dolomite 4.3(5) - - 2.3(5) 2.8(5) 2.2(5) 15.0(3) - - 4.6(2) - - K-feldspar 3.5(4) 4.2(3) 1.7(3) 2.9(4) 5.1(4) 2.1(3) ------magnetite ------4.9(1) 2.5(1) 3.6(2) 4.7(1) 1.3(1) 4.2(2) muscovite/Illite 9.6(8) 10.9(5) 15.8(5) 8.3(8) 13.1(1) 8.4(7) 2.2(6) 10.3(5) 7.8(6) 9.3(6) 20.4(4) 2.2(7) plagioclase - 22.4(3) 3.3(4) - - - - 6.3(3) - - - - quartz 2.0(3) 32.6(2) 28.8(2) 3.8 (4) 5.4(2) 4.1(1) 3.5(2) 17.6(1) 5.6(1) 10.2(2) 29.1(2) 3.4(2) talc 0.5(2) - 0.5(2) - - - 0.5(2) 2.8(2) 2.1(2) 2.7(3) 3.3(3) 2.1(3) tremolite 7.6(6) 4.6(3) 1.8(4) 3.2(5) 7.0(5) 2.4(5) - 10.3(1) 3.9(5) - - - TOT 100 100 100 100 100 100 100 100 100 100 100 100 1138 1139 1140 1141
51
1142 1143
C13V2 C14V2 C15V2 E1 E2 E3 E4 E5 E6 E7 E8
Rwp (%) 8.29 8.61 7.6 9.44 10.9 10.44 9.23 10.31 8.36 6.6 6.01 Rp (%) 6.29 6.73 5.77 7.07 8.13 7.64 6.94 7.53 6.26 5.09 4.67 χ2 3.99 3.54 2.87 6.32 8.68 8.89 6.35 10.41 5.87 3.32 3.20 antigorite 35.4(5) 35.1(4) 6.3(4) 44.7(6) 21.4(5) 13.5(3) 7.1(2) 17.1(3) 3.3(2) 7.7(3) 8.8(3) brucite - - - 0.3(1) 2.0(3) 1.0(3) 1.1(2) 1.4(3) - - - calcite 12.0(3) 9.8(2) 5.0(2) 2.6(2) - 3.5(2) 2.9(2) 4.9(3) 2.0(5) 1.3(2) - chrysotile 14.7(5) 21.9(4) 5.2(3) 25.1(9) 39.5(9) 35.0(9) 39.5(8) 36.4(9) 15.2(9) 4.6(4) 3.3(5) clinochlore 6.7(4) 18.4(5) 4.1(5) 15.8(4) 25.4(7) 23.8(8) 32.6(6) 22.6(7) 23.5(8) 12.4(5) 20.0(5) dolomite 5.9(3) 5.8(2) - - - 1.4(2) - - 0.4(4) - - hematite - - - 1.3(2) 1.2(2) 1.5(2) 1.7(2) 1.8(2) 2.6(3) - - enstatite - - - 1.2(3) 1.9(4) 3.1(6) 2.5(7) 1.4(4) 10.5(9) 8.3(5) 6.0(5) forsterite - - - 4.0(6) 3.2(7) 6.4(7) 5.7(4) 9.3(6) 6.0(7) - - K-feldspar ------5.6(5) 5.5(4) magnetite 4.6(1) 3.5(1) - 5.0(2) 5.4(2) 6.2(2) 6.9(2) 5.1(2) - - - muscovite/illite 6.8(7) 2.3(6) 15.0(5) - - - - - 7.6(9) 13.9(9) 13.3(9) plagioclase 3.7(4) - 18.7(4) - - - - - 8.3(4) 12.9(4) 11.3(4) quartz 5.5(2) 1.9(1) 40.1(1) - - - - - 6.3(2) 12.5(2) 10.4(2) talc - 1.3(3) 1.6(2) - - 4.6(9) - - 8.4(9) 9.6(4) 9.0(4) tremolite 4.7(4) - 4.0(3) - - - - - 5.9(5) 11.2(3) 12.4(3) TOT 100 100 100 100 100 100 100 100 100 100 100 1144 1145
52
1146 TABLE 5. Summary of the calculated concentration of chrysotile asbestos in the
1147 four samples selected for the cross-check analysis.
Chrysotile Chrysotile Chrysotile Chrysotile Normative Method wt% wt% wt% wt% reference STD1 STD2 C2 E1 Weighed Weighed value value 5.11 10.44 DTA - 7.6 12.2 8.9 22.2 Italian D.M. FTIR 4.9 13.8 6.3 32.9 09-06-94 Italian D.M. PCOM 3.7 1.4 n.d. 7.5 09-06-94 Italian D.M. SEM 09-06-94 - 4.6 13.1 4.7 27.5 LAB1 Italian D.M. SEM 09-06-94 - 1.9 0.6 5.9 26.4 LAB2 XRPD external Italian D.M. n.d. n.d. n.d. n.d. standard 09-06-94 method XRPD Rietveld - 5.8 11.2 6.3 25.3 method 1148
1149
53